Beginner-Friendly Topic
Take your time - it's perfectly normal to re-read this topic 2-3 times. Try the interactive code editor below to run code yourself. Use the Q&A section to check your understanding before moving on. You've got this! 🚀
Types of AI
AI can be categorized in different ways: by how it learns (supervised, unsupervised, reinforcement) or by what it does (discriminative vs generative). Understanding these categories helps you understand why different AI systems work differently.
Three Main Learning Types
The most useful way to categorize AI is by how it learns:
Supervised Learning: The AI is given labeled examples. It learns the relationship between inputs and correct answers.
Unsupervised Learning: The AI is given data without labels. It finds hidden patterns and groups on its own.
Reinforcement Learning: The AI learns by taking actions and receiving rewards or penalties. Like training a dog with treats.
Deep Learning ⊂ Machine Learning ⊂ Artificial Intelligence
Examples of Each Type
- Supervised: Email spam detection (trained on labeled spam/not-spam emails)
- Supervised: House price prediction (trained on prices of houses with known features)
- Unsupervised: Customer segmentation (grouping customers with similar behavior without predefined labels)
- Unsupervised: Anomaly detection (finding unusual patterns without knowing what unusual looks like in advance)
- Reinforcement: Game playing (AlphaGo learned to play Go by playing millions of games and winning rewards)
- Reinforcement: Robot movement (robot learns to walk by trying movements and getting reward when it stays upright)
Tip
Tip
Supervised learning is the most common type in real-world applications. Most AI you use every day was trained on labeled data. We will explore all three types in detail in Module 3.
Key Takeaways
- AI can be categorized in different ways: by how it learns (supervised, unsupervised, reinforcement) or by what it does (discriminative vs generative).
- Supervised: Email spam detection (trained on labeled spam/not-spam emails)
- Supervised: House price prediction (trained on prices of houses with known features)
- Unsupervised: Customer segmentation (grouping customers with similar behavior without predefined labels)